Lahjty

Use this before creating an ad from the user's real product or logo: imports exactly one image, either a url (for example a product photo URL from lahjty_extract_website_context, or a product's main photo URL read from a store connector the user connected) or a chat attachment (file), into the user's private Lahjty storage for 24 hours. role reference is the actual product or scene; role logo is an isolated brand mark. Decide the role from what the user said and what the image shows; ask if unclear, never from the filename alone. Returns a storage_id for the image, edit and ad tools. Free: no AI generation. Only use images the user supplied or approved.

lahjty_upload_ad_reference

Use this when the user asks how many credits they have left, or before a large batch to confirm it is affordable. Returns the balance and recent usage. Free.

lahjty_get_usage

Use this when the user asks what something will cost, or to confirm a planned request is complete before spending: returns the exact credit cost and any missing inputs without generating or reserving credits. Pass the same fields as the intended tool, without idempotency_key. Authorization uses the selected operation's own scope (copy:write, images:write, speech:write, campaigns:write or ad-library:write); a missing scope returns an insufficient_scope challenge so the host can ask the user to reconnect. If the user says to quote and wait, stop after this.

lahjty_validate_and_quote

Use this when the user has existing Arabic copy and wants it in another Arabic dialect (for example Egyptian to Saudi) with the same meaning, claims and formatting. Arabic output only. Uses copy credits. Do not use it to translate English; use lahjty_generate_marketing_copy with output_language ar instead.

lahjty_convert_arabic_dialect

Use this when the user wants a coordinated set in one step: captions plus an ad image and optionally a voiceover from one brief. Enable only the deliverables the user asked for. Takes longer and costs the sum of its parts; state the cost first. For a single image or caption use the individual tools instead.

lahjty_generate_campaign

Use this to turn an approved script into a voiceover for a reel, story, TikTok or WhatsApp voice note, in Arabic dialects or English. Reads the text exactly; write or approve the script first. Choose delivery (reel_voiceover for spoken ads) and a voice_profile or voice from lahjty_get_capabilities. Uses speech credits and saves playable audio; show the audio in the chat.

lahjty_generate_speech

Use this after the user picks a Lahjty Ad Library layout or product-photo format and lahjty_get_ad shows its fields: fills template_fields with the user's confirmed product facts and copy, keeps the product photo (reference_storage_ids) and logo separate, and renders the finished ad. Product-photo formats require a product photo and keep its printed label; use their suggested_aspect_ratio unless the platform needs another shape (4:5 feed, 9:16 stories). Pass brand colors as palette when known. Uses image credits and saves the result; state the cost first. Do not invent claims or drop required photos. Show the final image in the chat.

lahjty_generate_ad

Use this to create a new social media ad image, ideally built around the user's real product photo (reference_storage_ids) and optional logo. Pick aspect_ratio from the platform and keep the image text-free unless the user wants the offer or headline inside the image (then set include_text, overlay_text and output_language); text-free images need no language question. Uses image credits and saves the result; state the cost first unless the user already approved it. For a curated layout use lahjty_search_ad_library instead. To change an image you already made, use lahjty_edit_image, not this tool. Show the returned image in the chat.

lahjty_generate_marketing_image

Use this when the user wants to change an image Lahjty already made in this conversation: background, lighting, layout, product size, colors, text in the image, a different product photo or adding the logo. For the same ad in another size (a story, feed or square version) with nothing else changed, send resize true and aspect_ratio without an instruction: the new size's words and product are checked like a new image. Keeps everything the user did not ask to change. Pass the image's request_id as source_request_id. Uses image edit credits and saves a new version; the original stays unchanged. Do not use it for a brand-new concept; use lahjty_generate_marketing_image. Show the edited image in the chat.

lahjty_edit_image

Use this when the user wants ideas, proven layouts or a curated product-photo format before creating, for example "show me ad styles for a perfume". Pass their request in brief, naming the product and its packaging (bottle, jar, box, bag). For a physical product the user will photograph, set kind product_photo_format: these ready formats build the ad around the real product photo, keep its printed label, need no preparation and list a suggested_aspect_ratio. Leave other filters unset unless the user insists; a cafe is sector food, never an invented industry. Results include preview images and exact ad IDs; show them as a gallery and let the user pick. Fit to the brief is not measured performance. If nothing matches, broaden the search yourself and say what you relaxed. Free. For an ad ID the user pasted, use lahjty_get_ad.

lahjty_search_ad_library